ClearML - Auto-Magical CI/CD to streamline your AI workload. Experiment Management, Data Management, Pipeline, Orchestration, Scheduling & Serving in one MLOps/LLMOps solution
always_create_from_code argument to PipelineController (default True) to allow remote execution to create the Pipeline DAG at runtime (as opposed to adhering to the existing structure stored on the task when running locally)OutputModel reporting and other methods fail if OutputModel.update_weights() is not called before that (#1078)--overrides supportspawn start method for Python multiprocess. This should help circumvent issues like this
LazyEvalWrapper type error (#1081)include_archive parameter to Dataset.list_datasets(), now possible to include archived datasets in the returned list (#1069, thanks @natephysics!)aws.boto3.multipart_chunksize and aws.boto3.multipart_threshold configuration options (#1059, thanks @cgaudreau-ubisoft!)PipelineController.get_pipeline() for retrieving previously run pipelinescontinue_last_task=0 is ignored in pipelines run with retry_on_failure (#1054)Task.connect_configuration() doesn’t handle dictionaries with special charactersTask.get_by_name() doesn't return the most recent task when multiple tasks have same nameboto3 certificate verification ignores custom self-signed certificateslightning>=2.0 (#1033, thanks @aweinmann!)clearml-init support for standard HTTP and HTTPS ports in webapp in conjunction with non-default api/files server ports (#1031, thanks @pktiuk!)jsonargparse configuration filesValueError when setting task properties at the start of a pipeline componentPath where str expecteddict added to task with Task.connect()
Task.connect race condition overwriting task description with connected input modelDataset.add_external_files() (#962, thanks @john-zielke-snkeos!)Task.launch_multi_node() for distributed experiment executionTask.get_all_reported_scalars() to fetch all available scalar dataTask.remove_input_models() to disassociate input models from a taskDataset.list_datasets() to include dataset version in the resultjoblib hangs (#1009)gs:// (#1018, thanks @pzarfos!)Task.report_text() sometimes reporting to an incorrect task when multiple tasks run simultaneouslyTask.set_offline(offline_mode=False) raising an UnboundLocalError
scikit-image package (skimage) is sometimes not detected as a dependencyPipelineDecorator sometimes causing a race condition when starting a remote executionGetAllRequest in Task docstrings (#982, thanks @Make42!)jsonargparse during remote execution (clearml-agent #153 and #1010)ProxyDictPostWrite.update() not triggering a write-back to the backend (#985)TypeError when using Task.query_tasks() without specifying the task_filter parameterTask.init(continue_last_task=0) when running remotely, now no longer ignoredurllib3 import error when using urllib3>=2.0.0
Task.query_tasks() using TaskTypes (#938 )PipelineController
Dataset.create() usage for old server API versionsTask.get_tasks docstring (#937)Task.get_reported_scalars() docstring for x_axis parameter when set to "timestamp" (#964, thanks @jday1!)APIClient.events.debug_images() for latest server API versioncast parameter to Task.get_parameter method (#958, thanks @harry-optimised!)APIClient().models.get_all and APIClient().tasks.get_all requestsModel and Task
StorageManager.download_file() and StorageManager.download_folder() will not create a subfolder with the bucket name in the generated download path (#709)Task.mark_completed, Task.close, and the hyperparameter example (#927, thanks @Make42!)Task.get_debug_samples() to retrieve task's debug samples (#761)artifact_{serialization,deserialization}_function parameters in PipelineController and PipelineDecorator (#689)Logger.report_table() through extra_data parameter, solving (#796)Model.report_* methodsforce_download argument to Model.get_local_copy() to force the downloading of a new copy of a model even if already in the cachePipelineController.connect_configuration() to add configuration dictionaries to a pipelineDataset.delete configurable hosted contents deletion support, allowing deletion of both ClearML file server hosted files, and external storage hosted filesPipelineController.is_successful() criteria customizationCLEARML_VCS_DIFF environment variable behavior, setting it to an empty string now forces the task to not log uncommitted changesoutput_uri=True argument in the hook config for the OpenMMLab example--local argumentsubdirectory argument when pip installing from a git repo (#947, thanks @jday1!)SyntaxError (#959, thanks @jday1!)clearml-init (clearml-server #181 and #910)task.connect() docstring (ClearML Docs issue #473)Dataset.finalize() raises exception due to backward compatibility issue (#908)gradio is not importedpython-fire supportTask.connect() inside PipelineDecorator
get_or_create_project crashes when run in paralleltask.get_user_properties(value_only=True), previously raising an exceptionsilence_errors parameter not working as expected in StorageManager.get_file_size_bytes()
Dataset metadata name to a string prefixed with data_ breaking Dataset.get